A method and system for identifying heavy-load vehicle loads in multiple bridge deck driving scenarios

Through the identification method and deep learning model of bridge deck multi-driving scenarios, combined with the axle coupling numerical simulation technology, the problem of low recognition accuracy of traditional bridge dynamic weighing systems in complex traffic environments is solved, and the accurate identification of heavy-loaded vehicles is achieved and improve system stability.

CN118711140BActive Publication Date: 2024-11-26NINGBO LANGDA ENG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411196142.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-29
Publication Date
2024-11-26
Estimated Expiration
2044-08-29

AI Technical Summary

Technical Problem

Traditional bridge dynamic weighing systems are difficult to accurately identify heavy-loaded vehicles in complex traffic environments, and the structural strain response is sensitive to local damage, affecting the recognition accuracy, and poor system stability and durability.

Method used

The heavy-duty vehicle load recognition method is adopted for multi-driving scenes of bridge decks. The driving status of heavy-duty vehicles is recognized in real time through the visual monitoring module and the displacement monitoring module. Combined with the axle coupling numerical simulation technology and deep learning model, the deflection inversion vehicle weight model is generated and optimized. Vehicle load identification.

Benefits of technology

In complex traffic environments, the precise identification of heavy-loaded vehicles is achieved, which reduces the impact of local structural damage on identification and improves the stability and service life of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118711140B_ABST
    Figure CN118711140B_ABST
Patent Text Reader

Abstract

The present application discloses a method and system for identifying the load of heavy-loaded vehicles in multiple driving scenarios on a bridge deck. The method includes the following steps: identifying the driving status of heavy-duty vehicles on the bridge deck; calculating the mid-span deflection of the bridge in real time, and intercepting the corresponding mid-span deflection of the bridge according to the driving status identification time of the heavy-duty vehicle; selecting different deflection inversion vehicle models according to the driving status of the heavy-duty vehicle to identify the vehicle weight parameters. The system is used to implement the above method. The beneficial effects of the present application: firstly determining the driving status of the target heavy vehicle through computer vision, and then calculating the vehicle weight using different vehicle weight identification models, which is suitable for accurate identification of the weight of the target heavy vehicle under complex vehicle driving conditions such as single vehicles and multiple vehicles. And by eliminating the weight of small vehicles traveling around the target heavy vehicle, the accuracy of vehicle weight identification can be further improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of transportation technology, and in particular to a method and system for identifying the load of heavy-loaded vehicles in multiple driving scenarios on a bridge deck. Background Art

[0002] The dynamic weighing system of vehicle load can be divided into the pavement dynamic weighing system (PWIM) and the bridge dynamic weighing system (BWIM). The pavement weighing system needs to block traffic, dig up the road surface at the selected location of the road, install weighing sensors, and then measure the axle weight of the moving vehicle. The disadvantages of this method are that it affects vehicle traffic, is inconvenient to construct, has poor durability, is difficult to maintain, and has high costs. The bridge dynamic weighing system usually installs sensors at the bottom of the bridge to collect structural responses (such as structural dynamic strain) and uses relevant identification algorithms (such as Moses algorithm) to calculate the axle weight of the moving vehicle. The disadvantage of this method is that the strain response of the bridge structure is more sensitive to the change of the local stiffness of the bridge. For example, after a heavy truck passes over the bridge, cracks are generated at the sensor position, causing the local stiffness of the position to decrease, making the vehicle-induced strain response measurement of the bridge structure larger, resulting in deviations in the axle weight calculation. The development and change of cracks will also cause the accuracy of subsequent vehicle axle weight identification. Summary of the invention

[0003] One of the objectives of the present application is to provide a method for identifying heavy-loaded vehicle loads in multiple bridge driving scenarios that can solve at least one of the defects in the above-mentioned background technology.

[0004] Another object of the present application is to provide a heavy-load vehicle load identification system that can solve at least one of the defects in the above-mentioned background technology.

[0005] In order to achieve at least one of the above purposes, the technical solution adopted by the present application is: a method for identifying the load of heavy-loaded vehicles in multiple driving scenarios on a bridge deck, comprising the following steps:

[0006] S100: Identify the driving status of heavy vehicles on the bridge deck, including single-vehicle driving and multi-vehicle driving;

[0007] S200: Calculate the mid-span deflection of the bridge in real time, and intercept the corresponding mid-span deflection of the bridge according to the driving state recognition time of the heavy vehicle;

[0008] S300: Select different deflection inversion vehicle models according to the driving state of the heavy vehicle to identify vehicle weight parameters.

[0009] Preferably, step S100 includes the following process:

[0010] S110: Identify vehicle information of all vehicles on the bridge deck, including vehicle type, lane, license plate, speed, and time of entering and leaving the bridge;

[0011] S120: Positioning and identifying the heavy vehicle traveling on the bridge deck based on the acquired vehicle information;

[0012] S130: Determine the driving state of the heavy vehicle according to the located driving trajectory of the heavy vehicle.

[0013] Preferably, step S110 includes the following specific processes:

[0014] S111: For a bridge with a single span shorter than a set length, visual identification of vehicles on the bridge deck is performed at the exit, entrance or mid-span of the bridge;

[0015] S112: For a bridge with a single span greater than a set length, visual identification of vehicles on the bridge deck is performed at both the entrance and exit ends of the bridge;

[0016] S113: Perform algorithm analysis based on the visual recognition data to obtain vehicle information detected in real time on the bridge deck.

[0017] Preferably, step S120 includes the following specific process:

[0018] S121: storing the vehicle information of all vehicles when entering the bridge in database A, and storing the vehicle information of all vehicles when leaving the bridge in database B;

[0019] S122: searching the vehicle information in the database A based on the vehicle model, screening out the heavy vehicles that meet the requirements, and extracting the vehicle model and license plate information of the heavy vehicles;

[0020] S123: searching for vehicle information of the same vehicle model and license plate from database B based on the filtered vehicle model and license plate information;

[0021] S124: The target vehicle's crossing time at the bridge end and the crossing time at the bridge end searched in step S123 are merged to obtain a database C;

[0022] The lanes on the bridge deck include undivided two-way lanes and divided one-way lanes; the vehicle information of the divided one-way lanes is stored in databases A0, B0 and C0; the vehicle information of the two directions of the undivided two-way lanes is stored in databases A1, B1 and C1, and databases A2, B2 and C2 respectively.

[0023] Preferably, the process of judging the single-vehicle driving state of a heavy vehicle is as follows: based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C0, query the heavy vehicle's bridge-entry crossing time information in database A0; or, based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C1, query the heavy vehicle's bridge-entry crossing time information in databases A1 and A2; if there is no information on other heavy vehicles except the target heavy vehicle, the target heavy vehicle is marked as a single-vehicle driving state.

[0024] Preferably, the driving states of multiple vehicles include following vehicles in the same direction, merging vehicles in the same direction, and meeting vehicles in different directions. The specific judgment process is as follows: based on the time when the target heavy vehicle enters the bridge end and crosses the bridge end in database C0, query the time information of the heavy vehicle entering the bridge end in database A0; or, based on the time when the target heavy vehicle enters the bridge end and crosses the bridge end in database C1 or C2, query the time information of the heavy vehicle entering the bridge end in databases A1 and A2; if only the heavy vehicle information exists in database A in the same direction as the target heavy vehicle, and the lane information is consistent with the target heavy vehicle, then the heavy vehicle information is obtained. If the heavy vehicle has the same information as the target heavy vehicle, or the lane information of the two vehicles is different but the time interval between the two vehicles crossing the bridge end exceeds the set time threshold t1, the queried heavy vehicle and the target heavy vehicle are judged to be following vehicles in the same direction; if only the heavy vehicle information in the same direction as the target heavy vehicle exists in database A, and the lane information is different from that of the target heavy vehicle, and at the same time, the time interval between the two vehicles crossing the bridge end is within the set time threshold t2, the queried heavy vehicle and the target heavy vehicle are merging vehicles in the same direction; if only the heavy vehicle information in the opposite direction to the target heavy vehicle exists in database A, the queried heavy vehicle and the target heavy vehicle are meeting vehicles in opposite directions.

[0025] Preferably, step S200 includes the following process:

[0026] S210: Acquire real-time displacement data of the bridge span and store it in a displacement database;

[0027] S220: intercepting displacement data of corresponding time nodes in the displacement database according to the identified time information of the heavy vehicle;

[0028] S230: Preprocessing the intercepted data, eliminating non-structural displacement and temperature deformation data in the data, and obtaining vehicle-induced deflection dynamic response data.

[0029] Preferably, step S300 includes the following process:

[0030] S310: Generate simulation training data of deflection inversion vehicle weight model based on vehicle-bridge coupling numerical simulation technology and different heavy vehicle driving conditions;

[0031] S320: constructing a single-vehicle deep learning model and a multi-vehicle deep learning model according to the driving conditions of different heavy vehicles, and using simulation training data to perform model training;

[0032] S330: Using the weighing system near the bridge to obtain vehicle load information and match the information of vehicles crossing the bridge, forming a small sample data set under different heavy vehicle driving conditions, and continuing to optimize the model trained in step S320 through transfer learning;

[0033] S340: Based on the driving state of the heavy vehicle identified on the bridge and the vehicle-induced deflection dynamic response data calculated under the bridge, a corresponding deflection inversion vehicle weight model is selected, and the vehicle-induced deflection dynamic response data is input into the model to calculate the vehicle weight of the heavy vehicle;

[0034] S350: Optimize the vehicle weight data of the heavy vehicle according to the number of other small vehicles identified during the time when the target heavy vehicle goes on and off the bridge and the average vehicle weight of the corresponding statistical distribution, and update the optimized data in the vehicle information.

[0035] A heavy-duty vehicle load identification system is used to implement the above-mentioned heavy-duty vehicle load identification method for multiple bridge deck driving scenarios, comprising a visual monitoring module, a displacement monitoring module and a computing terminal; the visual monitoring module is installed on the bridge deck to identify the heavy-duty vehicle driving parameters on the bridge deck, and the displacement monitoring module is installed at the bottom of the beam in the middle of the bridge span to detect the displacement data of the bridge; the computing terminal is respectively communicated with the visual monitoring module and the displacement monitoring module to respectively receive the vehicle driving parameters and the displacement data of the bridge span, and then invert the load data of the heavy-duty vehicle under different driving conditions through the received data.

[0036] Preferably, the visual monitoring module includes a high-definition camera, which is installed at the entrance and / or exit of the bridge by means of a pole or a vertical pole; the displacement monitoring module includes a target and a displacement camera; the target is fixedly installed at the bottom of the mid-span beam of the bridge, and the displacement camera is installed at the pier at the bottom of the monitoring span beam of the bridge, and the displacement camera obtains the displacement data of the bridge by capturing images of the target.

[0037] Compared with the prior art, the beneficial effects of this application are:

[0038] The driving state of the target heavy vehicle is first determined through computer vision, and then the vehicle weight is calculated using different vehicle weight recognition models. This is suitable for accurate recognition of the weight of the target heavy vehicle under complex vehicle driving conditions such as single vehicle and multiple vehicles. The weight of small vehicles traveling around the target heavy vehicle is eliminated to further improve the accuracy of vehicle weight recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 The figure is a schematic diagram of the overall workflow of this application.

[0040] Figure 2 This is a schematic diagram of a heavy-duty vehicle following a vehicle in the same direction as the previous one in this application.

[0041] Figure 3 This is a schematic diagram of heavy vehicles merging left and right in the same direction in this application.

[0042] Figure 4 This is a schematic diagram of heavy vehicles meeting each other from the opposite direction in this application. DETAILED DESCRIPTION

[0043] Below, the present application is further described in conjunction with specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form new embodiments.

[0044] In the description of the present application, it should be noted that directional words, such as the terms "center", "lateral", "longitudinal", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", etc., indicating directions and positional relationships are based on the directions or positional relationships shown in the accompanying drawings, and are only for the convenience of narrating the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and cannot be understood as limiting the specific scope of protection of the present application.

[0045] It should be noted that the terms "first", "second", etc. in the description and claims of the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0046] The terms "including" and "having" and any variations thereof in the specification and claims of the present application are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0047] One aspect of the present application provides a method for identifying heavy-load vehicle loads in multiple bridge deck driving scenarios, such as Figure 1 As shown, one of the preferred embodiments includes the following steps:

[0048] S100: Identify the driving status of heavy vehicles on the bridge deck, where the driving status includes single-vehicle driving and multi-vehicle driving.

[0049] S200: Calculate the mid-span deflection of the bridge in real time, and intercept the corresponding mid-span deflection of the bridge according to the driving state recognition time of the heavy vehicle.

[0050] S300: Select different deflection inversion vehicle models according to the driving state of the heavy vehicle to identify vehicle weight parameters.

[0051] Those skilled in the art should know that in order to solve the drawbacks mentioned in the background technology, the technical problems to be solved by this application are as follows: (1) The measurement of the strain response of the bridge structure is sensitive to local damage to the bridge structure, which affects the accuracy of the vehicle axle weight identification, and the stability and durability of the traditional weighing system used to measure the strain of the bridge structure are poor. (2) The vehicle driving state of the bridge during actual service is relatively complex. It is rare for a single heavy vehicle to pass through the bridge. There are also cases where multiple heavy vehicles pass through the bridge at the same time and there are multiple small vehicles passing through the bridge at the same time around a single heavy vehicle. When the span of the bridge exceeds the set length, other vehicles on the bridge deck at the same time will interfere with the bridge structure strain response corresponding to the target vehicle, resulting in a large error in the identified axle weight of the target vehicle.

[0052] Based on the above technical problems, this embodiment can identify the bridge deck after the heavy vehicle drives onto the bridge, and then judge the driving state of the heavy vehicle on the bridge deck. And in the process of the heavy vehicle driving on the bridge deck, the time point when the heavy vehicle generates the corresponding driving state can be determined, and then the actual strain response of the bridge structure when the driving state occurs can be located according to the determined time point, and then the strain response of the bridge structure can be analyzed and inverted in combination with the driving state of the heavy vehicle to obtain the required axle weight of the heavy vehicle. Compared with the traditional method, this embodiment can accurately identify the axle of the target heavy vehicle in a complex traffic driving environment.

[0053] In addition, the structural strain response of the bridge is calculated by the change in the mid-span deflection of the bridge. This indicator reflects the state of the overall stiffness of the structure. Local structural damage such as cracks has little effect on it. The measured vehicle-induced structural response is more accurate and not easily disturbed by other factors, making the monitoring results more accurate, and the required structure is simpler and has a longer service life. For easy understanding, the specific contents of each step will be described in detail below.

[0054] It is important to know that heavy vehicles include trucks, vans, tankers, construction vehicles, large passenger cars and buses.

[0055] In this embodiment, step S100 specifically includes the following process:

[0056] S110: Vehicle information of all vehicles on the bridge deck is identified, including vehicle type, lane, license plate, speed, and time of entering and exiting the bridge.

[0057] S120: Positioning and identifying heavy vehicles traveling on the bridge deck based on the acquired vehicle information.

[0058] S130: Determine the driving state of the heavy vehicle according to the located driving trajectory of the heavy vehicle.

[0059] It should be known that the identification of vehicle types is used to determine whether there are heavy vehicles among the vehicles traveling on the bridge deck; the identification of license plates is used to identify and locate heavy vehicles at different time points; the identification of lanes, vehicle speeds, and the time of entering and exiting the bridge are all used to determine the driving status of multiple heavy vehicles. There are many ways to identify vehicle information, generally using visual monitoring. The specific identification process will be described in detail in the following content.

[0060] It should also be known that the vehicle information obtained in step S110 includes small vehicles and heavy vehicles, so heavy vehicles can be screened and their driving trajectory on the bridge deck can be located based on the obtained vehicle information. Finally, analysis can be performed based on the obtained driving trajectory of the heavy vehicles. If there are no other heavy vehicles around the target heavy vehicle at a certain moment, the target heavy vehicle is in a single-vehicle driving state; if there are other heavy vehicles around the target heavy vehicle at a certain moment, the target heavy vehicle is in a multi-vehicle driving state; the specific driving scenarios of the multi-vehicle driving state will be described in the following content.

[0061] It can be understood that vehicle identification on the bridge deck is related to the actual length of the bridge, that is, the range of visual monitoring is limited. If the length of the bridge is too long, multi-point monitoring may be required. If the length of the bridge is short, only a single monitoring point is required. For ease of understanding, the specific process of step S110 can be described in detail below.

[0062] Specifically, step S110 includes the following specific processes:

[0063] S111: For a bridge with a single span less than a set length, visual identification of vehicles on the bridge deck is performed at the exit, entrance or mid-span of the bridge.

[0064] S112: For a bridge with a single span greater than a set length, visual identification of vehicles on the bridge deck is performed at both the entrance and exit ends of the bridge.

[0065] S113: Perform algorithm analysis based on the visual recognition data to obtain vehicle information detected in real time on the bridge deck.

[0066] It should be known that the overall length of the bridge corresponds to the length of a single span of the bridge. If the length of a single span of the bridge is shorter, the number of spans of the bridge is relatively small, and the overall length of the bridge is shorter. Conversely, the overall length of the bridge is longer. The set length of the single span for judging the length of the bridge can be selected according to the actual needs of those skilled in the art. For example, the single span of 50m can be used as the set length. A specific description is given below.

[0067] For bridges with a single span of less than 50m, the overall length is short, generally smaller than the visual recognition range of a single monitoring point; therefore, the entire bridge deck can be visually recognized at one end or the middle of the bridge. For bridges with a single span of more than 50m, the overall length is long, and generally multiple monitoring points are required for visual recognition; therefore, visual recognition can be performed at both ends of the bridge toward the middle of the bridge span, and the superposition of the visual recognition ranges of the two monitoring points can basically cover the entire bridge.

[0068] It should also be known that the data of visual recognition is image data. In order to obtain the vehicle license plate, lane, vehicle model, speed, and time of entering and exiting the bridge, the image data needs to be analyzed by a corresponding algorithm. The required vehicle information is obtained by extracting features from continuous image data. There are many specific algorithms required for image data processing, and the specific processing process is well known to those skilled in the art, so it will not be elaborated in detail here.

[0069] Specifically, the positioning of the heavy vehicle in step S120 includes the following specific processes:

[0070] S121: The vehicle information of all vehicles when entering the bridge is stored in database A, and the vehicle information when leaving the bridge is stored in database B.

[0071] S122: Search the vehicle information in the database A based on the vehicle model, screen out the heavy vehicles that meet the requirements, and extract the vehicle model and license plate information of the heavy vehicles.

[0072] S123: Searching for vehicle information of the same model and license plate from database B based on the extracted model and license plate information.

[0073] S124: The target vehicle's crossing time at the bridge end and the crossing time at the bridge end searched in step S123 are merged to obtain a database C.

[0074] It should be known that the above positioning process is only for positioning vehicles in one-way lanes, while the lane conditions of the bridge deck can generally include undivided two-way lanes and divided one-way lanes. For divided one-way lanes, since the driving directions of the vehicles are the same, there is only one database A, B and C for storing vehicle information; for the convenience of subsequent description, the databases storing vehicle information of divided one-way lanes can be marked as databases A0, B0 and C0. Similarly, for undivided two-way lanes, databases A, B and C for storing vehicle information are stored in each direction; for the convenience of distinction, the databases for storing vehicle information in the two directions can be marked as databases A1, B1 and C1, and databases A2, B2 and C2, respectively.

[0075] It is important to know that the judgment of the driving state of multiple heavy vehicles is related to the specific conditions of the lanes; the driving state of multiple vehicles mainly includes following vehicles in the same direction, parallel driving in the same direction, and meeting vehicles in the opposite direction. For a divided one-way lane, the possible driving states of heavy vehicles include single vehicle driving, following vehicles in the same direction, and parallel driving in the same direction. For an undivided two-way lane, the possible driving states of heavy vehicles include single vehicle driving, following vehicles in the same direction, parallel driving in the same direction, and meeting vehicles in the opposite direction. For ease of understanding, the following will describe in detail the judgment process of the specific driving state of heavy vehicles in combination with the lane conditions.

[0076] 1. Judgment on the single-vehicle driving of heavy vehicles.

[0077] (1) For a bridge with divided one-way lanes, based on the time when the target heavy vehicle enters the bridge and crosses the bridge, the time when the heavy vehicle enters the bridge in database A0 is queried. If there is no information about other heavy vehicles except the target heavy vehicle within the set time range, the target heavy vehicle can be marked as a single vehicle.

[0078] (2) For bridges without two-way lanes, based on the target heavy vehicle’s crossing time at the bridge end and the crossing time at the bridge end in database C1, query the heavy vehicle’s crossing time at the bridge end in databases A1 and A2. If there is no information on other heavy vehicles except the target heavy vehicle, the target heavy vehicle is marked as a single vehicle.

[0079] 2. Determine the heavy vehicles that are following in front and behind.

[0080] (1) For bridges with divided one-way lanes; Figure 2As shown, based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C0, the heavy vehicle's bridge-entry crossing time information in database A0 is queried. If the heavy vehicle information exists in database A0 and is the same as the target heavy vehicle's lane information, or if the two have different lane information but the interval between the bridge-entry crossing time exceeds the set time threshold t1, it can be determined that the queried heavy vehicle and the target heavy vehicle are the same front and rear following vehicles.

[0081] (2) For bridges without two-way lanes; Figure 2 As shown, based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C1, the heavy vehicle's bridge-entry crossing time information in databases A1 and A2 is queried. If only database A1 has the heavy vehicle information, and it has the same lane information as the target heavy vehicle, or the two have different lane information but the interval between the bridge-entry crossing time exceeds the set time threshold t1, then it can be determined that the queried heavy vehicle and the target heavy vehicle are the same front and rear following vehicles.

[0082] Of course, the heavy vehicle crossing time information of the bridge end can also be queried in databases A1 and A2 based on the target heavy vehicle crossing time of the bridge end in database C2. If only database A2 has the heavy vehicle information, and the lane information is the same as that of the target heavy vehicle, or the lane information of the two is different but the interval between the crossing time of the bridge end exceeds the set time threshold t1, then it can be determined that the queried heavy vehicle and the target heavy vehicle are the same front and rear following vehicles.

[0083] It should be known that, for the target heavy vehicle and the queried heavy vehicle driving in the same direction but different lanes, they may be driving in the same direction, following each other, or driving side by side in the same direction. Therefore, for further judgment, the time interval between the two vehicles entering the bridge can be compared with the set threshold value t1. If the time interval between the two vehicles is less than the threshold value t1, it means that the two vehicles basically enter the bridge in a flush posture, that is, they are in the driving state of driving side by side in the same direction. If the time interval between the two vehicles is greater than the threshold value t1, it means that the two vehicles enter the bridge in a front-to-back posture, that is, they are in the driving state of driving side by side in the same direction.

[0084] The value of the threshold t1 can be determined according to actual needs. The value of the threshold t1 is generally related to the vehicle speed. For example, the minimum distance between two heavy vehicles following each other can be set to half the length of the vehicle body. Generally speaking, the length of a heavy vehicle body is 6-12m. Taking the middle value of 8m as an example, the minimum distance between vehicles is 4m. Assuming the vehicle speed is 60km / h, the time required for the vehicle to travel 4m is 0.24s, that is, the value of the threshold t1 is 0.24s.

[0085] 3. Judgment on the left and right merging of heavy vehicles in the same direction.

[0086] (1) For bridges with divided one-way lanes; Figure 3 As shown, based on the time of the target heavy vehicle entering the bridge end and the time of the target heavy vehicle leaving the bridge end in database C0, the time of the heavy vehicle entering the bridge end in database A0 is queried. If the heavy vehicle information exists in database A0, and the lane information is different from that of the target heavy vehicle, and the interval between the time of the heavy vehicle entering the bridge end and the time of the target heavy vehicle leaving the bridge end is within the set time threshold t2, it can be determined that the queried heavy vehicle and the target heavy vehicle are merging in the same direction.

[0087] (2) For bridges without two-way lanes; Figure 3 As shown, based on the time of the target heavy vehicle entering the bridge end and the time of the target heavy vehicle leaving the bridge end in database C1, the time of the heavy vehicle entering the bridge end in databases A1 and A2 is queried. If only database A1 has the information of the heavy vehicle, and the lane information is different from that of the target heavy vehicle, and the interval between the time of the heavy vehicle entering the bridge end and the time of the target heavy vehicle leaving the bridge end is within the set time threshold t2, it can be determined that the queried heavy vehicle and the target heavy vehicle are merging in the same direction.

[0088] Of course, the heavy vehicle crossing time information of the bridge end can also be queried in databases A1 and A2 based on the target heavy vehicle crossing time of the bridge end in database C2. If only database A2 has the heavy vehicle information, and the lane information is different from that of the target heavy vehicle, and the interval between the crossing time of the bridge end is within the set time threshold t2, then it can be determined that the queried heavy vehicle and the target heavy vehicle are parallel vehicles in the same direction.

[0089] It should be known that if the interval between the target heavy vehicle and the queried heavy vehicle entering the bridge end and crossing the line exceeds the set time threshold t2, it can be considered that the target heavy vehicle and the queried heavy vehicle are traveling in the same direction. Therefore, the value of threshold t2 can be the same as threshold t1, or less than the value of threshold t1.

[0090] 4. Judgment on heavy vehicles approaching from the opposite direction.

[0091] This driving state only occurs on bridges with undivided two-way lanes; Figure 4 As shown, based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C1, the heavy vehicle's bridge-entry crossing time information in databases A1 and A2 is queried. If only database A2 has heavy vehicle information, it can be determined that the queried heavy vehicle and the target heavy vehicle are meeting in opposite directions.

[0092] Of course, the heavy vehicle crossing time information in databases A1 and A2 can also be queried based on the target heavy vehicle crossing time at the bridge end and the target heavy vehicle crossing time at the bridge end in database C2. If only database A1 has the heavy vehicle information, it can be determined that the queried heavy vehicle and the target heavy vehicle are meeting in the opposite direction.

[0093] In this embodiment, Figure 1 As shown, step S200 includes the following process:

[0094] S210: Acquire real-time displacement data of the bridge mid-span and store it in a displacement database.

[0095] S220: intercepting the displacement data of the corresponding time node in the displacement database according to the identified time information of the heavy vehicle.

[0096] S230: Preprocessing the intercepted data, eliminating non-structural displacement and temperature deformation data in the data, and obtaining vehicle-induced deflection dynamic response data.

[0097] It should be known that there are many ways to obtain the displacement data of the bridge span, which can be achieved through visual recognition or distance recognition. The displacement deformation of the bridge may be caused by vehicle driving, temperature changes, air flow, etc. Therefore, after obtaining the displacement data of the bridge, it is necessary to eliminate the displacement deformation data not caused by vehicle driving, so that the final remaining bridge displacement data is vehicle-induced data, and then the required vehicle-induced deflection dynamic response data can be obtained through corresponding calculations.

[0098] In this embodiment, Figure 1 As shown, step S300 includes the following process:

[0099] S310: Generate simulation training data of the deflection inversion vehicle weight model based on the vehicle-bridge coupling numerical simulation technology and different heavy vehicle driving conditions.

[0100] S320: According to the driving conditions of different heavy vehicles, a single-vehicle deep learning model and a multi-vehicle deep learning model are constructed respectively, and the model training is performed using the simulation training data in step S310.

[0101] S330: Use the weighing system near the bridge to obtain vehicle load information and match the information of vehicles crossing the bridge to form a small sample data set under different heavy vehicle driving conditions, and continue to optimize the model trained in step S320 through transfer learning.

[0102] S340: Based on the driving state of the heavy vehicle identified on the bridge and the dynamic response data of the vehicle-induced deflection calculated under the bridge, a corresponding deflection inversion vehicle weight model is selected, and the dynamic response data of the vehicle-induced deflection is input into the model to calculate the vehicle weight of the heavy vehicle.

[0103] S350: Optimize the vehicle weight data of the heavy vehicle according to the number of other small vehicles identified during the time when the target heavy vehicle goes on and off the bridge and the mean vehicle weight of the corresponding statistical distribution, and update the optimized data in the vehicle information databases C0, C1 and C2.

[0104] It should be known that the specific working principle of the vehicle-bridge coupling numerical simulation technology is a well-known technology for those skilled in the art, so it will not be elaborated in detail here; at the same time, the establishment process of the inverse vehicle weight model is also a well-known technology for those skilled in the art, so it will not be elaborated in detail here. The weighing system near the bridge can be a nearby weighbridge or a weighing system set at the entrance of the bridge.

[0105] Another aspect of the present application provides a heavy-load vehicle load identification system for implementing the above-mentioned heavy-load vehicle load identification method for multiple bridge deck driving scenarios, such as Figure 1 As shown, one preferred embodiment includes a visual monitoring module, a displacement monitoring module and a computing terminal. The visual monitoring module is installed on the bridge deck, and the visual monitoring module can identify the driving parameters of heavy vehicles on the bridge deck by machine vision. The displacement monitoring module is installed at the bottom of the bridge beam in the middle span, and the displacement monitoring module uses machine vision to detect and identify the displacement of the bridge. The computing terminal is respectively communicated with the visual monitoring module and the displacement monitoring module to receive the vehicle driving parameters and the displacement data of the bridge respectively, and then identify the driving state of the vehicle and calculate the mid-span deflection of the bridge through the corresponding algorithm, so as to invert the load data of the heavy vehicle under different driving states through the generated inversion vehicle weight model.

[0106] In this embodiment, there are multiple specific structures of the visual monitoring module. For ease of understanding, one of the structures will be described in detail below. The visual monitoring module includes a high-definition camera, which is installed at the entrance and / or exit of the bridge by means of a pole or a pole. The high-definition camera can send the collected data to the computing terminal, and then the computing terminal can generate the required vehicle information through a deep learning algorithm.

[0107] It should be known that there are many types of deep learning algorithms, which can specifically include target detection algorithms, target segmentation algorithms, and character recognition algorithms; the specific algorithm working principle is a well-known technology for those skilled in the art, so it will not be elaborated in detail here. At the same time, the specific structure and working principle of the high-definition camera are well-known technologies for those skilled in the art, so it will not be elaborated in detail here. The number of high-definition cameras installed is related to the length of the bridge. For the convenience of understanding, a specific example will be used to explain it in detail below.

[0108] Specifically, for a bridge with a single span shorter than a set length, a high-definition camera can be installed at the exit end of the bridge in each direction by using a pole or a vertical pole; for a bridge with a single span longer than a set length, a high-definition camera can be installed at the entrance end and exit end of the bridge in each direction by using a pole or a vertical pole. The high-definition camera can be connected to the computing terminal for communication, and the image data captured by the high-definition camera can be sent to the computing terminal, so that the computing terminal can use an algorithm to analyze the image data and perform real-time detection of vehicle information.

[0109] In this embodiment, there are multiple specific structures of the displacement monitoring module. For ease of understanding, one of the structures will be described in detail below. The displacement monitoring module includes a target and a displacement camera; the target can be fixedly installed at the bottom of the mid-span beam of the bridge by artificial means, and the target can move synchronously with the displacement deformation of the bridge. The displacement deformation of the bridge can be detected more conveniently through the setting of the target. The displacement camera is installed on the pier at the bottom of the monitoring span beam of the bridge and the shooting direction is aimed at the target. The displacement camera captures the image of the target and sends the collected data to the computing terminal. The computing terminal can then calculate through image processing algorithms and other related technologies to obtain the original displacement data of the target image and store it in the displacement database. Then, the displacement data of the target image in the displacement database is intercepted according to the time information of the identified heavy vehicle, and the data is preprocessed to eliminate the non-structural displacement and temperature deformation data in the data, and finally the required vehicle-induced deflection dynamic response data is obtained.

[0110] It should be known that there are many types of image processing algorithms, and the working principles are well-known to those skilled in the art, so they will not be elaborated in detail here. The specific structure and working principle of the displacement camera are also well-known to those skilled in the art, so they will not be elaborated in detail here. Compared with the sensor detection method used in the traditional bridge structure deflection collection, this embodiment can effectively improve the accuracy and stability of the collection results through visual monitoring; at the same time, according to the installation position of the displacement camera, the service life can be effectively improved.

[0111] The above describes the basic principles, main features and advantages of the present application. Those skilled in the art should understand that the present application is not limited by the above embodiments, and the above embodiments and the specification only describe the principles of the present application. The present application may have various changes and improvements without departing from the spirit and scope of the present application, and these changes and improvements fall within the scope of the present application for which protection is sought. The scope of protection claimed by the present application is defined by the attached claims and their equivalents.

Claims

1. A method for identifying heavy-duty vehicle loads in multiple bridge deck driving scenarios, characterized in that: The steps include: S100: Identify the driving status of heavy vehicles on the bridge deck, including single-vehicle driving and multi-vehicle driving; S200: Calculate the mid-span deflection of the bridge in real time, and intercept the corresponding mid-span deflection of the bridge according to the driving state recognition time of the heavy vehicle; S300: selecting different deflection inversion vehicle models according to the driving state of the heavy vehicle to identify vehicle weight parameters; Step S300 includes the following process: S310: Generate simulation training data of deflection inversion vehicle weight model based on vehicle-bridge coupling numerical simulation technology and different heavy vehicle driving conditions; S320: constructing a single-vehicle deep learning model and a multi-vehicle deep learning model according to the driving conditions of different heavy vehicles, and using simulation training data to perform model training; S330: Using the weighing system near the bridge to obtain vehicle load information and match the information of vehicles crossing the bridge, forming a small sample data set under different heavy vehicle driving conditions, and continuing to optimize the model trained in step S320 through transfer learning; S340: Based on the driving state of the heavy vehicle identified on the bridge and the dynamic response data of the vehicle-induced deflection calculated under the bridge, a corresponding deflection inversion vehicle weight model is selected, and the dynamic response data of the vehicle-induced deflection is input into the model to calculate the vehicle weight of the heavy vehicle; S350: Optimize the vehicle weight data of the heavy vehicle according to the number of other small vehicles identified during the time when the target heavy vehicle goes on and off the bridge and the average vehicle weight of the corresponding statistical distribution, and update the optimized data in the vehicle information.

2. The method for identifying heavy-loaded vehicles on a bridge deck in multiple driving scenarios according to claim 1, characterized in that: Step S100 includes the following process: S110: Identify vehicle information of all vehicles on the bridge deck, including vehicle type, lane, license plate, speed, and time of entering and leaving the bridge; S120: Positioning and identifying the heavy vehicle traveling on the bridge deck based on the acquired vehicle information; S130: Determine the driving state of the heavy vehicle according to the located driving trajectory of the heavy vehicle.

3. The method for identifying heavy-duty vehicle loads in multiple bridge deck driving scenarios according to claim 2, characterized in that: Step S110 includes the following specific processes: S111: For a bridge with a single span shorter than a set length, visual identification of vehicles on the bridge deck is performed at the exit, entrance or mid-span of the bridge; S112: For a bridge with a single span greater than a set length, visual identification of vehicles on the bridge deck is performed at both the entrance and exit ends of the bridge; S113: Perform algorithm analysis based on the visual recognition data to obtain vehicle information detected in real time on the bridge deck.

4. The method for identifying heavy-loaded vehicle loads in multiple bridge deck driving scenarios as claimed in claim 2, characterized in that: Step S120 includes the following specific processes: S121: storing the vehicle information of all vehicles when entering the bridge in database A, and storing the vehicle information of all vehicles when leaving the bridge in database B; S122: searching the vehicle information in the database A based on the vehicle model, screening out the heavy vehicles that meet the requirements, and extracting the vehicle model and license plate information of the heavy vehicles; S123: searching for vehicle information of the same vehicle model and license plate from database B based on the filtered vehicle model and license plate information; S124: The target vehicle's crossing time at the bridge end and the crossing time at the bridge end searched in step S123 are merged to obtain a database C; The lanes on the bridge deck include undivided two-way lanes and divided one-way lanes; The vehicle information of the divided one-way lane is stored in databases A0, B0 and C0; the vehicle information of the two directions of the undivided two-way lane is stored in databases A1, B1 and C1, and databases A2, B2 and C2 respectively.

5. The method for identifying heavy-loaded vehicles on a bridge deck in multiple driving scenarios according to claim 4 is characterized in that: The process of judging the driving status of a single heavy vehicle is as follows: Based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C0, query the heavy vehicle's bridge-entry crossing time information in database A0; Alternatively, based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C1, query the heavy vehicle's bridge-entry crossing time information in databases A1 and A2; If there is no information about other heavy vehicles except the target heavy vehicle, the target heavy vehicle is marked as a single vehicle driving state.

6. The method for identifying heavy-loaded vehicles on a bridge deck in multiple driving scenarios according to claim 4, characterized in that: The driving status of multiple vehicles includes vehicles following in front and behind, vehicles merging in the same direction, and vehicles meeting in different directions. The specific judgment process is as follows: Based on the target heavy vehicle's bridge-entry crossing time and bridge-exit crossing time in database C0, query the heavy vehicle's bridge-entry crossing time information in database A0; Alternatively, based on the time of the target heavy vehicle entering the bridge end and crossing the bridge end in database C1 or C2, query the time of the heavy vehicle entering the bridge end and crossing the bridge end in database A1 and A2; If only the heavy vehicle information exists in the database A in the same direction as the target heavy vehicle, and the lane information is the same as the target heavy vehicle, or the lane information of the two is different but the interval between the crossing time of the bridge end exceeds the set time threshold t1, the queried heavy vehicle and the target heavy vehicle are judged to be the same forward and rearward vehicles; If only the heavy vehicle information in the database A in the same direction as the target heavy vehicle exists, and the lane information is different from that of the target heavy vehicle, and the interval between the two crossing the line at the bridge end is within the set time threshold t2, it is determined that the queried heavy vehicle and the target heavy vehicle are merging in the same direction; If only the heavy vehicle in the opposite direction to the target heavy vehicle exists in the database A, it is determined that the queried heavy vehicle and the target heavy vehicle are meeting in the opposite direction.

7. The method for identifying heavy-loaded vehicle loads in multiple bridge deck driving scenarios according to any one of claims 2 to 6, characterized in that: Step S200 includes the following process: S210: Acquire real-time displacement data of the bridge span and store it in a displacement database; S220: intercepting displacement data of corresponding time nodes in the displacement database according to the identified time information of the heavy vehicle; S230: Preprocessing the intercepted data, eliminating non-structural displacement and temperature deformation data in the data, and obtaining vehicle-induced deflection dynamic response data.

8. A heavy-duty vehicle load identification system, used to implement the heavy-duty vehicle load identification method for multiple bridge deck driving scenarios according to any one of claims 1 to 7, characterized in that: include: Visual monitoring module; The visual monitoring module is installed on the bridge deck to identify the driving parameters of heavy vehicles on the bridge deck; Displacement monitoring module; The displacement monitoring module is installed at the bottom of the bridge beam in the middle of the span to detect the displacement data of the bridge; Computing terminal; the computing terminal is respectively connected to the visual monitoring module and the displacement monitoring module to receive the vehicle driving parameters and the displacement data of the bridge span, and then invert the load data of the heavy vehicle under different driving conditions through the received data.

9. The heavy-duty vehicle load identification system according to claim 8, characterized in that: The visual monitoring module includes a high-definition camera; the high-definition camera is installed at the entrance and / or exit of the bridge by means of a pole or a vertical pole; The displacement monitoring module includes a target and a displacement camera; the target is fixedly installed at the bottom of the mid-span beam of the bridge, and the displacement camera is installed at the pier at the bottom of the monitoring span beam of the bridge. The displacement camera acquires the displacement data of the bridge by collecting images of the target.